Eternitysound - First seak peak at Chemical Kinetics: Tranistion Samples

[GeorgeZ]

Thanks kbaccki and tcollins. Your input has been invaluable. It's a lot to take on board, so I'll mull it over a bit before I come up with some more problems errrr, questions I mean

For what it's worth though, below is a little rough demo track I've made, not dissimilar to my synth patch bank demos. This is by no means anything final, but will serve as a good indication as to the direction I'm heading with these. Any feedback will be appreciated!

http://eternitysound.com/mp3s/Chemical% ... 0Demo1.mp3

[james_mcfadyen]

Many people will not even bother downloading a free library, since they believe it's a waste of time.

Really!? I've had over 700 downloads of my Cool Rhodes library in just over 2 weeks.

[kbaccki]

Thanks kbaccki and tcollins. Your input has been invaluable. It's a lot to take on board, so I'll mull it over a bit before I come up with some more problems errrr, questions I mean

For what it's worth though, below is a little rough demo track I've made, not dissimilar to my synth patch bank demos. This is by no means anything final, but will serve as a good indication as to the direction I'm heading with these. Any feedback will be appreciated!

http://eternitysound.com/mp3s/Chemical% ... 0Demo1.mp3

Hey thanks for the demo... have to admit, I was getting a little curious. Definately have some intricate waveforms there... And it sounds fantastic. Nice complex sweeps, and useful for more than just straight music production, I think.

RE: r8brain free file clobbering... I downloaded r8brain free, and it appears to have a batch option that allows you to rename the original w/ a prefix and/or suffix. It looks like your original wav's would get renamed to *.wav.orig by default if no options are changed. In addition, there's an explicit option on the batch setup page that requires the user to explcitly delete the source file. So I'm thinking if both prefix and suffix are blank and the delete source is not checked, perhaps the original is untouched and a companion *_r8b.wav is created...

RE: file size difference between waveosaur and r8brain... First, I poked around the net to refresh my memory on some of the trickiness w/ the WAV format. Because the format is really so open-ended (e.g., custom file headers may cause some apps to not be able to read a WAV file if they see stuff they're not expecting, or not prepared for, etc.), I was thinking there must be a valid reason for the difference... moreso, is one or the other 24-bit formats preferred for compatibility sake?

My understanding is that a WAV file reading/writing app could deal with 24-bit sample (WAV PCM) data as either "word aligned" (i.e., samples aligned at 16-bit boundaries), or simply byte aligned. In the case of word-aligned, you would see no file size difference between 32-bit float format and 24-bit int format -- both cause each sample "container" (as MS calls it) to contain 4 bytes of data. In the 24-bit case the file reader and/or app would simply know to ignore the unused least significant byte (which should be 0 anyway). In the case of byte-aligned, each "container" contains 3 bytes of data. The 3-vs-4 bytes per container is possible because the WAV format header allows you to specify data alignment & bits-per-sample resolution independently. The file reader/app needs to deal with all possibilities, again because the WAV format is very flexible in that regard. In fact, you could store 5.1 channel (interleaved) 20-bit resolution data in 3-byte sample "containers" if you wanted... as long as the data chunk is byte aligned, which basically comes down to adding a padding byte at the end to make the whoe data chunk evenly divisible by 8-bits... the padding byte is not considered part of the data proper and ignored.

Why would one choose word-aligned (4-byte) 24-bit int data, when all you need is 3-bytes of data per sample? Simply for efficiency... CPUs deal with 16-bits at a time minimally... and of course modern CPUs deal w/ 64-bits, 128-bits, etc. at a time... but no less than 16-bits. If you "pack" your 24-bit int data into 3-byte containers, you simply need to spend some CPU on the backend to unpack it into 16-bit aligned units. In practice... the extra byte per sample for 4-byte data format vs. 3-byte format data is... uh... divide by... carry the one... 25% more disk space. For streaming that data from disk in realtime (via sampler or DAW), maybe you get some savings? On a relatively modern computer, a modest 25% overhead in disk space for, say, a 100MB sample lib is a drop in the storage bucket... that said, the CPU savings you get from having 16-bit aligned sample data read directly from disk compared to 8-bit aligned is basically so close to 0% that it's not worth even coming up with a number. Maybe 10+ years ago, unpacking those bits could have been measurable in the 1-2% CPU range, but I don't think so today...

In any event, I would say if you don't have to support transmitting extra unused bits over the interweb, then all the better... in that case a lossless compression format, like RAR, will effectively compress your 24-bit samples down to 18-bits (or less!) of information anyway... depending on the sample content, of course... So storage and distribution will never be an issue on your end...

So anyway... that's why you may see file size differences between apps. E.g., from what I can tell, in CoolEdit (Adobe Audition), even if you save out as 24-bit file w/ 20-bit/sample resolution, you get a file that is 16-bit aligned... so basically the same file size as the 32-bit float you're sourcing from...

I would say just be careful w/ Waveosaur, Goldwave, WaveLab, or whatever you use that they're not adding their own wacky WAV headers that cause the files to not be 99.9% compatible with all other apps... Know what I mean?

[Kriminal]

Going by the Vengeance standard of pricing (which is a relative average figure of course, and they are VERY known) that would be $20 for 150 samples... that seems awfully expensive to me! What do you think?

Vengeance are selling a name/brand (and sometimes other ppls samples ) so you cant cmpare their prices to yours

sell them for what you think they are worth, time is money.

[GeorgeZ]

Updated the first post with a little demo. Have a listen and tell me if I'm on the right track Cheers!

[GeorgeZ]

Bump?